HIGH MLflow: MLflow tracking UI/API reachable without authentication
Why it matters
An open MLflow server lets strangers list experiments, read artifact paths, and often pull logged models or credentials stored as params.
Fix it — 3 steps
- Enable MLflow basic auth or put the tracking server behind an authenticating reverse proxy with TLS.
- Bind the tracking server to 127.0.0.1 if it is only used locally.
- Block the port at the firewall: `sudo ufw deny 5000`.
Verify it’s fixed
curl -m 5 http://YOUR_SERVER_IP:5000/ # should not return the MLflow UI to strangers
References
Not sure if your stack is exposed?
Run the free check — 30 seconds, safe read-only probes.
Scan your stack